Sixty-eight patients with proven esophageal varices were studied at Kenyatta National Hospital, Nairobi, Kenya. Of these patients, 29.4% had schistosomal portal hypertension, 22.1% cirrhosis and only 8.8% extrahepatic portal vein occlusion. One quarter of the patients had a normal liver biopsy and extrahepatic portal vein that was demonstrated to be patent. Problems relating to liver biopsy sampling resulting in underdiagnosis of specific causes of esophageal varices such as schistosomiasis are discussed. We argue that many of these patients were likely to be suffering from idiopathic portal hypertension, a condition apparently not previously recognized in Africa. Of this last group, 70.6% had suffered gastrointestinal bleeding, as had 50% of the patients with schistosomiasis. Together these two groups accounted for three-quarters of all patients who had bled. The detection of eggs of Schistosoma mansoni in stool and/or rectal snip correlated well with liver biopsy findings in both a positive and negative sense. Only 18% of patients with negative stools and snips had evidence of schistosomiasis in the liver, and positive stools or snips were found in only 14.6% of patients without schistosomal liver involvement. Of the patients in the study, 50% were of the Kamba tribe, although only 12.9% of all medical admissions to the hospital were Kamba (P < 0.01). Luo patients were significantly more frequent within the group with schistosomiasis (P < 0.02). Esophageal varices were attributed to tropical splenomegaly syndrome in only one patient. The implications of our results are discussed and our findings are compared with previous work from East Africa.